Cell lines
The Human Neural Stem Cells (NSCs) were obtained from GIBCO. They are derived from the NIH approved H9 (WA09) human embryonic stem cell line. They retain their normal female human karyotype and the potential to differentiate into neurons and glial cell populations following multiple passages.
NSCs were maintained and amplified up to 4 passages using the NSCs media (KO DMEM/F12; StemPro Neural Supplement 2%, Glutamax 1%, bFGF 20 ng/mL, EGF 20 ng/mL) in matrigel coated plates.
Neuronal differentiation
The human NSCs (Gibco) (passage < 5) were seeded in matrigel coated p100 Petri dishes, maintained in NSC culture media (KO DMEM/F12, StemPro Neural Supplement 2%, Glutamax 1%, bFGF 20 ng/mL, EGF 20 ng/mL) and changed every other day until confluence. After confluence, media was changed to Neuro Progenitor induction Media (NPM) (DMEM/F12, B27 1%, N2 0.5%, Glutamax 1%) for 7 days, changing the media completely every other day. Then cells were washed once with PBS, detached with accutase (5–10 min), resuspended in DMEM and centrifuged at 250 g. Supernatant was discarded and cells resuspended in Neuronal Optimized Media complete (NOMc) (DMEM/F12, B27 2%, N2 1%, laminin 1 µg/mL, cAMP 100 nM, ascorbic acid 200 ng/mL, BDNF 10 ng/mL, GDNF 10 ng/mL, IGF 10 ng/mL). Depending on the experiment, different densities of progenitors were seeded in matrigel coated surface (DIV0) and maintained in NOMc for neuronal maturation. Unless stated, experiments were performed at DIV40 of neuronal maturation.
Chamber assembly and neuronal growth
The microfluidic chambers (Xona microfluidics) were cleaned with ethanol and the coverslips coated overnight with a poly-ornithine solution (0.1 mg/mL in PBS). Before bonding, poly-ornithine from coverslips and ethanol from chambers were both removed by thorough washes with distilled water. Bonding was performed by placing the microfluidic device over the dry coverslip and applying a uniform pressure on the upper surface of the microfluidic chamber as detailed in the manufacturer’s instructions (Xona microfluidics). After assembling, the reservoirs of the microfluidic chamber were filled with PBS, avoiding to generate any bubble in the channel system. PBS was removed (not completely to avoid bubbles) and matrigel added in all of the reservoirs to completely coat both compartments. Chambers were maintained in the incubator at 37 °C for at least 1 h. Immediately before seeding, matrigel was removed from the chambers and NPCs (cf. Neuronal differentiation) seeded in the top well of the somatic compartment (300,000 cells in 50 µL of NOMc). Cells flowed from the top well through the somatic compartment where they attached. Chambers were then placed into the incubator for 30 min, after which the wells were topped up with NOMc. Media levels in the wells of the somatic compartment should be kept higher than the levels of the axonal compartment. Cultures were maintained by changing media completely once per week and equilibrated three days after. After approximately 20 days in culture, we observed axons starting to cross the flow channel.
Injury protocol
The syringe pump (NE-1002X, New Era Pump Systems Inc) was fitted with a syringe (HSW, Norm-Ject 12 mL) and connected to a tubing (GE Healthcare peek tubing green 0.75 mm i.d., 1/16 o.d.; #18-1112-53) entering the microscope plexiglass incubator. The syringe and the tubing were loaded with PBS and purged in a way that liquid continuity was not interrupted by any bubble. Microfluidic chambers were placed in the confocal microscope stage (Zeiss Confocal LSM780, Zeiss Live LSM7) preheated to 37 °C and equilibrated to 5% CO2. Depending on the experiment, medium in the microfluidic chamber was completely replaced by aCSF (NaCl 121 mM, KCl 2.5 mM, CaCl2 2.2 mM, MgSO4 1 mM, NaHCO3 29 mM, NaH2PO4 0.45 mM, Na2HPO4 0.5 mM, glucose 20 mM, pH = 7.4) prior to the experiment. The tubing from the syringe pump was inserted into the opening of the flow channel of the microfluidic chamber and a basal flow in the withdrawal direction (1–10 µL/min) maintained until the injury was performed. For the injury, a 100 µL/min flow rate was applied for 90 s in the withdrawal direction, after which it was immediately reversed for 10 s and returned to the basal flow for the time required by the experiment.
Speed measurements and analysis
3–3.4 µm fluorescent beads (BioLegend) were resuspended in PBS and 80 µL of the solution loaded to the middle bottom well. All the other wells were topped up with PBS. The system was mounted in the microscope stage and starting from basal flow of 10 µL/min, different pump flow rates (30, 50, 100, 150 and 200 µL/min) were selected in the syringe pump. Using brightfield and epifluorescence illumination, movies were recorded (@800 fps) in the flow channel for 90 s. In each movie the speed of 10 beads was measured by kymograph plots every 10 s interval. The continuity equation was used to estimate the speed in the flow channel:
$$A_{sy} \times V_{sy} = A_{ch} \times V_{ch} .$$
where A: area, V: velocity, sy: syringe, ch: channel.
A simplified version of the microfluidic chamber geometry has been analyzed using commercial Finite Elements Analysis software suite (Comsol Multiphysics 5.2) to calculate velocity profile and shear stress at the walls.
Force measurements and analysis
For measuring the stress at different pump flow rates, we used a urethane microstamp matrix printed with round pillars of 5 µm diameter: 10 µm tall: 12 µm spacing (RMS microstamps). The microstamp was embedded in OCT and cut in a cryostat into 40 µm sections. With the help of a scalpel and a stereomicroscope, the resulting sections were further cut to obtain pieces of approximately 50 µm in length, 30 µm in height (including pillars height 10 µm) and 40 µm in depth. One piece per microfluidic chamber was placed in a way that the pillars were facing to the side (perpendicular to the flow channel long axis). In this position, when placed in the microscope, the pillars could be seen from the side, making it easier to visualize the bending. Microfluidic chambers were bonded to the coverslip and filled with PBS. The chamber was connected to the syringe pump and placed in the stage of the microscope. A basal flow rate of 10 µL/min was maintained before applying a 400 µL/min flow rate for 90 s. Movies were recorded and bending of the windward pillar measured. Maximal force acting on the pillar was calculated based on Hooke’s law [65]:
$$F = \frac{{3 \cdot \pi \cdot E_{u} \cdot D^{4} }}{{64 \cdot L^{3} }} \cdot \delta$$
where F is the force applied to the pillar, Eu is the Young’s modulus of urethane (1.38 MPa), D is the diameter of the pillar (5 µm), L is the length of the pillar (10 µm) and δ is the displacement of the center of the pillar.
For further calculating the stress of the axon we used:
$$\upsigma = {\text{F/A}}$$
where σ is stress, F is force, and A is area of the axon (typical axon of length 60 µm and 1 µm diameter).
RT-PCR
Total RNA was isolated from cultures at various stages of differentiation (NSC, NPC and mature neurons) using the RNeasy Plus Micro kit (Qiagen) according to the manufacturer’s protocol. Samples for each stage were taken for NSCs stage at confluence, for NPC stage at the end of the progenitors differentiation (7 days in NPM media) and neuronal stage at DIV40 of the neuronal maturation (cfr. Neuronal differentiation). HEK cells were used as the control. Reverse transcription was performed with the Transcriptor First Strand cDNA Synthesis Kit (Roche) in a two-step RT-PCR. Briefly, 250 ng of RNA were mixed with 1 μL of Anchored-oligo (dT)18 primer and water to a final volume of 13 μL. The denaturation step was performed at 65 °C for 10 min. Samples were then cooled on ice and 7 μL of RT mix were added. RT conditions were: 10 min at 25 °C, 30 min at 48 °C, and 5 min at 95 °C. For the qPCR, 1 μL of the 10 × cDNA dilution was used. qPCR reactions were performed in triplicate with 1 μL of cDNA and 9 µL of Power SYBR® Green PCR Master Mix (Applied Biosystem) in a final volume of 10 µL using the LightCycler® 480 Instrument II (Roche) under the following cycling conditions: initial denaturation at 95 °C for 3 min, 50 cycles of 30 s at 95 °C, 30 s at 58 °C and 45 s at 72 °C, and a final step of 1 min at 72 °C. Data were analysed to obtain the ΔCT per target gene. Data from five independent experiments were pooled and normalised against the three housekeeping genes (RPL13, RPL27 and β2M). The sequences of the primer pairs for the specific target genes were:
genes
|
Fwd (5ʹ–3ʹ)
|
Rev (5ʹ–3ʹ)
|
---|
β2M
|
CTCGCGCTACTCTCTCTTTCTG
|
GCTTACATGTCTCGATCCCACT
|
RPL13
|
GGACCTCTGTGTATTTGTCAATTTT
|
GCTGGAAGTACCAGGCAGTG
|
RPL27
|
ACATTGATGATGGCACCTCAG
|
CCAAGGGGATATCCACAGAGT
|
GFAP
|
AGGAAGATTGAGTCGCTGGAG
|
CGGTGAGGTCTGGCTTGG
|
MAP2
|
ATTCCGAGGTTCCAACACAC
|
ACCAGCCATTGAAGAAATGC
|
OLIG2
|
CGGCTTTCCTCTATTTTGGTT
|
GTTACACGGCAGACGCTACA
|
PAX6
|
CAGGTGTCCAACGATGTG
|
GTCGCTACTCTCGGTTTACTAC
|
SOX2
|
GCACATGAACGGCTGGAGCAACG
|
TGCTGCGAGTAGGACATGCTGTAGG
|
ß3-Tub
|
TGGGCGACTCGGACTTGC
|
CCACTCTGACCAAAGATGAAATTG
|
Flow cytometry
Cells were analysed from cultures at different stages of differentiation (NSC, NPC and mature neurons). Samples for each stage were taken, for NSCs stage when they were at confluence, for NPC stage at the end of the progenitors differentiation (7 days in NPM media) and neuronal stage at DIV40 of the neuronal maturation (cfr. Neuronal differentiation). Briefly, cells for each sample were cultured in a well of 6 MW plate, maintained and differentiated up to the moment of harvesting for immunostainings. Cells were detached by incubating them in accutase and counted (approximately 250.000 cells/well). For surface markers (Cocktail A: CD15-488, CD24-APC, CD44-PE, CD184-PE/Cy, Cocktail B: CD56-450, CD140-PE/Cy and O4-488), the cells were pelleted, resuspended in FACS staining buffer (FSB) (0.5% Fetal Bovine Serum, 2 mM EDTA in PBS), followed by incubation with conjugated antibodies for 30 min on ice. Cells were then washed and pelleted twice with FACS Washing Solution (FWS) discarding the supernatant after every wash. Finally, the cells were resuspended in 200 µL of PBS and analysed on a FACS CANTO. For staining of intracellular markers (Cocktail C: Nestin-PE, GFAP-488, TUJ1-647, Cocktail D: SOX2-488 and Ki67-PE; and independently NeuN or Map2), cells were fixed with the Fixation buffer (FB) (eBioscience) for 40 min in the dark. After fixation, permeabilization buffer (PB) (eBioscience) was added to each tube and cells pelleted for 5 min at 250 g. After discarding the supernatant, cells were washed and pelleted one more time. Cells were then incubated with antibodies dissolved in 100 µL of PB for 60 min in the dark at RT. The cells were then washed once with PB, once with FWS, resuspended in PBS and analysed. For NeuN and Map2 after the wash they were incubated with anti-mouse AF488 for 1 h, washed and analysed. Unstained cell were used as negative control. The FACS CANTO raw data was analysed using FlowJo software.
Multielectrode array
NPCs were seeded in a 48 MW Multi-Electrode Array (MEA) plate (Axion) at 10.000 cells per well, and differentiated to neurons in NOMc as detailed elsewhere. Activity in each well was recorded at 37 °C for 2 min, every day for 12 weeks (Axion Maestro Middleman). Raw data was processed using the neural metrics tool (AxiS 2.4 software) and exported for further analysis. The general conditions set for the identification of activity were: a) an electrode is active and considered for analysis if #spikes > 1/min, b) spikes were detected when the amplitude was 6.5 SD higher than the noise average, c) a burst (train) consist of a minimum of 5 spikes with a maximum inter-spike interval of 100 ms. Four parameters were taken into account to describe the culture activity and maturation: the mean firing rate of active electrodes per well (Weighted Mean Firing Rate); Number of active electrodes per well; Number of bursts per well; and measure of synchrony between electrodes in a well (Synchrony Index). Three independent experiments with a minimum of four wells each were analyzed.
Transduction
Neurons grown on microfluidic chambers were transduced with lentiviral particles (LV-mCherry-Mem, LV-GFP, LV-APP-GFP, LV-SYP-GFP, LV-Mito-GFP). Particles were resuspended in NOMc media and after withdrawing the culture media in the somatic compartment, 200 µL were added to the top well. 14–16 h later the transduction media was discarded and replaced with fresh NOMc. After 5 to 7 days of transduction, expression was strong enough for conducting the experiments.
ELISA quantification
Samples were collected immediately after injury from the flow channel and the axonal compartment of the microfluidic chambers. Non-injured chambers were used as a control. 5–7 chambers were pooled per replicate in order to collect enough volume for the measurements. The proteins were extracted by applying 50 µL of RIPA 1X buffer supplemented with phosphatase and proteinase inhibitors to the axonal compartment well. Cells were scraped with the pipette tip and passed through a 0.8 syringe. Samples were incubated on ice for 30 min before a 20 min centrifugation step at 4 °C, 14.000 RPM.
The protein concentration was assessed via BCA assay (Thermo Scientific™ Pierce™ BCA Protein Assay Kit Catalogue Number: 23227) according to the manufacturer instructions. The concentration between samples was adjusted in order to have the same starting concentration in all the samples.
Samples underwent Enzyme-linked Immunosorbent Assay for quantitative detection of Human Aβ40 and Aβ42 according to the plate manufacturer instructions (Novus Biologicals ELISA kit Human Aβ1-40 cat. NBP2-69,909 and Human Aβ1-42 cat. NBP2-69913).
Briefly, samples and reagents were brought to RT. The standard curve was prepared by making serial dilutions of the standard provided with the kit in duplicate. 100 µL of standards and samples were loaded per well. The plate was incubated for 90 min, at 37 °C in the dark followed by incubation with 100 µL of biotinylated antibody for 1 h at 37 °C in the dark. The plate was then washed 4 times before incubation with HRP-conjugated secondary antibody for 30 min at 37 °C in the dark. 5 washes were performed after the incubation before applying the substrate reagent for 10 min and then the stop solution. The Absorbance was read at 450 nm.
Based on the Absorbance, the concentration in each well was calculated and plotted. Ratios between Aβ42 and Aβ40 were considered to address the impact on Amyloid Beta production in injured axons versus control.
Proteomics
We performed label free quantitative mass spectrometry in two independent experiments. The first one was a discovery experiment, which consisted in a proteomic analysis, the second one was a confirmatory experiment which consisted in proteomic and phosphoproteomic analysis. Protein extraction protocol was performed in the same way for both experiments. The remaining steps were performed in different facilities as detailed.
Protein extraction
Microfluidic chambers of DIV40 were segregated into control and injured group. For each sample, 6–7 chambers were pooled to obtain enough material. Control chambers were subjected to a 10 µL/min flow during 150 s, while injured ones were subjected to 10 µL/min 30 s, 100 µL/min 90 s and 30 µL/min 30 s. Immediately after, chambers were placed on ice, the media was removed completely from all the reservoirs and washed with PBS. 100 µL of lysis buffer (RIPA (10x) Buffer, SDS (10%) 1%, protease inhibitor 1%, phosphatase inhibitor 1%, dH2O was added to the top axonal reservoir and flushed trough the axonal compartment to the lower reservoir (repeated 4 times with pipette) to collect all the material from the axonal compartment. The chamber was checked under microscope for efficient collection of axons from the axonal compartment and flow channel. The same procedure was followed for the neuronal compartment. The collection of the first chamber was used to collect 2 more chambers, to avoid dilution of the sample. To disrupt membranes we aspirated the lysate with an insulin syringe for 20 times on ice. Tubes were kept on ice for 30 min and centrifuged at 12,000 g for 20 min at 4 °C. Finally, supernatant was collected, a small volume was used to quantify protein concentration by BCA, and the rest of the sample kept at − 80 °C.
Discovery experiment
Digestion
Samples were reduced with 10 mM dithiothreitol at 60 °C and Alkylated with 50 mM iodoacetamide at RT in the dark. Next, samples were digested with sequencing grade trypsin at 37 °C for 16 h. Digestion was quenched and the peptides collected in formic acid for application to RP-HPLC column. 1 µg of samples was injected to Orbitrap Fusion Lumos and LC–MS/MS data was acquired using designed LC protocol.
Mass spectrometry analysis
Trypsin-digested peptides were analyzed by ultra-high pressure liquid chromatography (UPLC) coupled with tandem mass spectroscopy (LC–MS/MS) using nano-spray ionization. The nanospray ionization experiments were performed using an Orbitrap fusion Lumos hybrid mass spectrometer (Thermo) interfaced with nano-scale reversed-phase UPLC (Thermo Dionex UltiMate™ 3000 RSLC nano System). Sample were loaded onto precolumn (volume less than 15 µL, Thermo ACCLAIM PepMap 100 P/N:164564-CMD) at 7 µL/min rate for 5 min followed by an analytical run using a 25 cm, 75 µm ID glass capillary packed with 1.7 µm C18 (130) BEHTM beads (Waters corporation). Peptides were eluted from the C18 column into the mass spectrometer using a linear gradient (5–100%) of ACN (Acetonitrile) at a flow rate of 395 μL/min for 2 h. The buffers used to create the ACN gradient were: Buffer A (98% H2O, 2% ACN, 0.1% formic acid) and Buffer B (80% ACN, 0.1% formic acid). Mass spectrometer parameters were as follows: an MS1 survey scan using the orbitrap detector (mass range (m/z): 400–1500 (using quadrupole isolation), 60,000 resolution setting, spray voltage of 2400 V, Ion transfer tube temperature of 290 °C, AGC target of 400,000, and maximum injection time of 50 ms) were followed by data dependent scans (top speed for most intense ions, with charge state set to only include + 2–5 ions, and 20 s exclusion time, while selecting ions with minimal intensities of 50,000 at in which the collision event was carried out in the high energy collision cell (HCD Collision Energy of 30%), and the fragment masses where analyzed in the ion trap mass analyzer (with ion trap scan rate of turbo, first mass m/z was 100, AGC Target 5000 and maximum injection time of 35 ms). Protein identification was carried out using Peaks Studio 8.5 (Bioinformatics solutions Inc.).
Data analysis
Data was analyzed using Peaks Studio 8.5. The data search was performed against UNIPROT_Human_9606. The search parameters were set as follows: Search Engine Name: PEAKS; Parent Mass Error Tolerance: 10.0 ppm; Fragment Mass Error Tolerance: 0.4 Da; Precursor Mass Search Type: monoisotopic; Enzyme: Trypsin; Max Missed Cleavages: 3; Digest Mode: Semispecific; Fixed Modifications: Carbamidomethylation: 57.02; Max Variable PTM Per Peptide: 3; Database: Human Uniprot; Taxon: All; Contaminant Database: contaminants; Searched Entry: 21,067; FDR Estimation: Enabled; Merge Options: no merge; Precursor Options: corrected; Charge Options: no correction; Filter Charge: 2–8; Process: true; Associate chimera: yes. The instrument parameters were: Fractions: IP1013_02.raw; Ion Source: ESI(nano-spray); Fragmentation Mode: high energy CID (y and b ions); MS Scan Mode: FT-ICR/Orbitra MS/MS;Scan Mode: Linear Ion Trap.
Confirmatory experiment
Digestion
Protein samples (~ 225 µg) representative of each of the experimental conditions, in four replicates per condition were digested using the FASP protocol [66] with some modifications. Briefly, samples were loaded and buffer exchanged in a 10 kDa MWCO filter unit, reduced with 20 mM tris(2-carboxyethyl)phosphine (TCEP) and alkylated with 100 mM dimethyl acrylamide (DMA). Proteins were digested overnight at 37 °C with Trypsin/Lys-C mix (Mass Spec grade, Promega) at 1:35 enzyme/substrate and shaken at 600 rpm. After tryptic digestion, peptides were collected and extracted with 5% formic acid (FA). Peptide samples were cleaned up using a C18 SPE cartridge (SupelCleanC18, Sigma-Aldrich) and vacuum dried in a Speed Vac. A fraction (~ 25 μg) was kept for direct LC‐MS analysis, and samples stored at– 20 °C.
Phosphopeptide enrichment
Phosphorylated peptides were enriched using 5 μL Fe(III)-NTA cartridges and a micro-syringe pump. Dried tryptic peptides ~ 200 μg were dissolved with 30 μL of 0.5% TFA and 120 μL of ACN just before loading the sample to the AssayMAP Fe(III)-NTA cartridges. The cartridges were conditioned with 100 μL 50% ACN, 0.1% TFA at a flow rate of 1200 μL/h and equilibrated with 50 μL loading buffer (80% ACN, 0.1% TFA) at 1200 μL/h. After conditioning, samples were loaded at 350 μL/h onto the cartridge. The cartridge was washed with 50 μL loading buffer follow by a second wash with 50 μL 50% ACN, 0.1% TFA. Phosphorylated peptides were eluted directly into 15 μL of net formic acid in a new 1.5 mL Eppendorf tube with 30 μL of Na2HPO4, 30 μl of 5% NH4OH and 30 μL 5% pyrrolidine, respectively. Pooled elution was cleaned up using a C18 OMIX SPE tip (Agilent, Milford, MA) and dried under low pressure and stored at − 20 °C.
Mass spectrometry analysis
Nano-LC MS/MS analysis was performed using an on-line system consisting of a nano-pump UltiMate™ 3000 UHPLC binary HPLC system (Dionex, ThermoFisher) coupled with an Orbitrap Fusion™ Lumos™ Tribrid™ mass spectrometer (ThermoFisher Scientific, Germany) fitted with an EASY-Spray™ nano-electrospray ion source. Whole cell extract peptides and phosphopeptides were resuspended in 1.6% ACN, 0.1% formic acid and injected directly into an EASY-Spray™ PepMap RSLC C18 capillary column (75 µm × 50 cm, 100 Å, 2 μm particle sizes) operated at 50 °C. Peptides were eluted into the MS, at a flow rate of 250 nL/min, using a 136 min gradient as follows: 2% B to 40% B in 120 min followed by ramping to 95% B in 11 min and washing for 5 min. Mobile phase A was 0.1% formic acid in H2O and mobile phase B was 80% acetonitrile and 0.1% formic acid. The mass spectrometer was operated in data-dependent mode, with a single MS scan in the Orbitrap (400–1600 m/z at 60 000 resolution (at 200 m/z) in a profile mode) followed by MS/MS scans in the Orbitrap on the 10 most intense ions at 15,000 resolution. Ions selected for MS/MS scan were fragmented using higher energy collision dissociation (HCD) at normalized collision energy of 27 with an isolation window of 1.4 Th.
Data analysis
Mass spectrometry raw data were processed using MaxQuant [67] version 1.6.1.0 with default settings. Dimethyl-Propionamide (C) was set as fixed modification, and acetyl (Protein N-term) and oxidation (M) were allowed as variable modification. The search was performed against the human UniProt database (95,128 sequences, released on October 2020). Protein quantification was based on two or more peptides using the LFQ approach [68]. Proteins were reported at 1% FDR.
MS raw files were loaded to Progenesis LC–MS software (version 4.1, Nonlinear Dynamics, UK) for label free quantification and analysis. Profile data of the MS scans were transformed to peak lists with respective peak m/z values, intensities, abundances and m/z width. MS/MS spectra were treated similarly. For retention time alignment of the samples, the most complex sample was selected as a reference, and the retention times of the other samples were aligned automatically to a maximal overlay of all features. Features with only one charge or more than four charges were excluded from further analyses. Raw abundances of the remaining features were normalized to allow correction for factors resulting from experimental variation.
Rank 1–5 MS/MS spectra were exported as Mascot generic file and used for peptide identification with MASCOT Version 2.4 (Matrix Science Ltd, UK) in the human protein database (Uniprot, 95,128 sequences, released on October 2020). Search parameters were peptide mass tolerance of 10 ppm, and MS/MS tolerance of 0.05 amu allowing 2 missed cleavage. Dimethyl-Propionamide of cysteine was set as a fixed modification, and Oxidation (M), Acetyl (Protein N-term), Phospho (ST) and Phospho (Y) were allowed as variable modification. Peptide assignments with an ion score cut-off of 30 and a significance threshold of p < 0.05 were re-imported to Progenesis. After summing up the ion intensities of all the peptides assigned to each protein, a list of proteins abundances was generated. One-way ANOVA was used to calculate the ρ-value based on the abundance values. Results were grouped according to the treatment condition.
Mass spectrometry data pre-processing
Quality control analyses including correlations, principal components analysis peptide counts and intensities were completed with Perseus (v 1.6.5.0). For the protein abundance analysis one complete experimental set was excluded for the discovery experiment and 2 experimental units (neuronal fraction control and injury) were excluded from the confirmatory experiment, based on outliers in intensities, protein numbers and principal component analysis. Proteins that were positive for “potential contaminant”, “reverse” and “only identified by site” were excluded, as well as a set of proteins that were present in the culture media (i.e., B27/N2 supplements) or extracellular matrix. For each fraction/treatment only proteins that had ≥ 2 peptides identified in at least two replicas were kept (Additional file 10: Table S1). For phosphopeptides only the ones with a score higher than 30 were kept.
PCA, Heatmaps and Volcano were processed in LFQ-Analyst [69] (Fig. 4A, B and Additional file 9: Fig. S3A–C). MaxQuant files were used as input, adjusted p-value cut-off was set to 0.05 and log2 fold change cutoff to 0.585. Perseus-type method was used as missing values imputation method and Benjamini Hochberg for FDR correction.
For phosphoproteomics, peptides with score > 30 were kept. Fold change was calculated with the ratio of the mean normalized abundance for control and injury. P-values and fold change for each phosphopeptide were used for Volcano Plots drawing.
Enrichment analysis
All enrichment analysis were implemented using g:Profiler [70]. For proteomic analysis two sets of proteins were used: the axonal and the neuronal set. Both of them consisted of the addition of the control and injury treatments set of proteins pre-processed as detailed previously (Additional file 10: Table S1). Over-representation analysis of human tissue specificity was performed using the neuronal protein set as input on the Human Protein Atlas database with a significance threshold of 0.05 and using the g:SCS algorithm for multiple testing correction (Fig. 4C and Additional file 9: Fig. S3D). Gene Ontology Biological Process database was used for functional enrichment analysis for the axonal and neuronal protein set independently with significance threshold of 0.05 and g:SCS for multiple corrections (Fig. 4D). To understand the enrichment of the axonal fraction versus the neuronal fraction, the axonal list of proteins was used as input query and the neuronal one as the background and assessed against the Gene Ontology database. Parameters were set at 0.05 for significance threshold and electronic GO annotations were excluded. The resulting significant terms for Cellular Compartment, Biological Process and Molecular Function were further reduced by exclusion of redundant terms using Revigo [71]. For generating the stringplots the lists of terms for each category were loaded in Cytoscape [72] with all the terms represented as nodes (p < 0.05) and the similarity between terms represented as edges (cutoff = 0.3) using EnrichmentMap plugin. Clustering was performed using the AutoAnnotate plugin, with MCL clustering algorithm and labeling for most present words in the nodes (Fig. 4E).
For phosphoproteomics analysis two sets of proteins were considered for each fraction. One set consisted in all the proteins that presented at least one identified phosphopeptide, the other is a subset with proteins that presented at least one phosphopeptide that showed significant changes in abundances between the treatments (p < 0.05, fold change > 1.5 or < 0.67). To understand the enrichment of the subset of proteins that significantly change their phosphorylation, this subset was used as input query and the set of all proteins with at least one protein identified was used as the background and assessed against the Gene Ontology Biological Process database (Fig. 5D). This was performed for both the axonal and the neuronal fractions. and assessed against the Gene Ontology database.
Also the subset of proteins that change their phosphorylation in the axonal fraction was used for functional enrichment analysis in the Gene Ontology Molecular Function database and the resulting list reduced by exclusion of redundant terms with Revigo (Fig. 5F).
Kinase prediction
Specific phosphorylation sites of the axonal protein set with their respective fold changes and significance values were used as the input for the KSEA App [73] with both the PhosphoSitePlus and NetworKIN databases. Three parameters of each kinase were followed: The background-adjusted value of the mean log2(fold change) of all the kinase’s substrates, the number of substrates for each kinase and the statistic significance of the score for each kinase. Kinases with less than two substrates were excluded (Additional file 12: Table S3). Plots were implemented using Coral [74] and Prism.
Immunofluorescence
Immunofluorescence of the neuronal culture in the microfluidic chambers was performed by first aspirating the media from the wells followed by one wash with PBS and immediate fixing in 4% paraformaldehyde (PFA) in PBS for 45 min at RT. PBS was added to all the wells and the chambers gently peeled off from the coverslip to minimize cell and axonal detachment. Coverslips were sectioned with a diamond pen to isolate the region of interest, and continued with blocking step as next detailed. For immunofluorescence of cells on coverslips, cells were washed with PBS and fixed with 4% PFA in PBS for 30 min at RT, followed by two PBS washes of 10 min each. To avoid non-specific staining, cells were incubated at RT for one hour with blocking solution (10% goat serum, 0.1% Triton X-100 in PBS). Cells were then stained with primary antibodies, dissolved in antibody solution (0.1% Triton X-100 in PBS) and incubated ON at 4 °C. Afterwards, cells were washed twice with PBS and stained with secondary antibodies in antibody solution at RT for 2 h. Finally, cells were stained for 5 min with DAPI and mounted on slides with Mowiol.
Fixed cells were examined with an inverted Zeiss LSM 780 confocal microscope (Zeiss, Germany) using an oil immersion objective (63X/1.4 NA). Images were acquired and processed with Zen Blue and ImageJ softwares, respectively.
WGA staining
Wheat Germ Agglutinin-Alexa Fluor 633 (Invitrogen) was diluted in PBS (10 µg/mL) and chambers incubated with the solution for 20 min. After one wash with PBS, cells were returned to the growth media for 30 min and imaged.
Mitochondrial membrane potential
After 5 days of transduction with Mito-GFP, neurons were incubated with tetramethylrhodamine ethyl ester perchlorate (TMRE) (100 nM) diluted in NOM for 30 min. After one wash with PBS they were incubated in aCSF and movies acquired during injury. TMRE intensity was measured during the whole period of the injury and normalized to the intensity of the first 10 frames (F0).
Non-specific membrane permeability
Neurons grown in microfluidic chambers were incubated with a mix of dextranes of different sizes labeled with Cascade Blue (3 KD MW), Alexa Fluor 488 (10 KD MW) and Tetramethylrhodamine (40 KD MW). For this, the dextranes (50 µM in NOM) were loaded in the three lower wells of the microfluidic chambers (80 µL in each well). After injury (10 µL/min × 30 s, 100 µL/min × 90 s, 10 µL/min × 30 s) or control (10 µL/min × 150 s) or positive control treatment (10 µL/min × 150 s, media supplemented with 0.1 mM Triton X-100), chambers were washed three times with aCSF and axonal images taken (40X objective). Intensity of the three different channels inside the axon was measured and a ratio of the intensity inside the axon versus the intensity outside the axon (background) calculated for normalization.
Calcium transients measurement
NPCs were seeded in multiwell chambers (IBIDI) and differentiated to neurons for 10, 30 and 40 days. They were incubated with the intracellular Ca2+ sensor Fluo-4 AM (5 µM in NOMc) for 30 min. After one wash with PBS they were returned to NOMc for 30 min. For imaging, the media was replaced by aCSF. Time lapse movies were recorded at 2 fps for 60 s at 40X/1.3 NA at 480 nm excitation with confocal live module (Zeiss LSM7). In the movies, 10 random projections were selected and intensity analysed using ImageJ. The presence of transients was considered when the maximum intensity levels increased by 10% from the resting levels.
Axonal ionic levels
Neurons grown in microfluidic chambers were incubated with sodium-binding benzofuran isophthalate (SBFI) (5 µM), FURA-2AM (2 µM) or potassium-binding benzofuran isophthalate (PBFI) (5 µM) in NOM with 0.1% Pluronic for 60 min. Followed by a wash with aCSF, they where returned to NOMc for at least 15 min. Injury was performed in aCSF media. Chambers were mounted in a Zeiss AxioObserver 7 microscope equipped with a HE Fura 2 shift free (E) filter set (Ex BP 340/30, Ex BP 387/15) and imaged with a 40X/1.2 CApo objective. Movies were recorded at 2.5 fps for FURA-2AM and PBFI and at 3.1 fps for SBFI, and axonal intensities of 340 nm and 380 nm measured. After background subtraction for each channel and normalization to the levels of the first 5 frames, the 340/380 nm ratios were calculated.
For control experiments, neurons were grown in IBIDI chambers, incubated with ratiometric probes as described, and incubated in aCSF before imaging. During imaging a solution of KCl (50 mM in aCSF) was perfused.
Measurements of real-time axonal changes
Movies of axons subjected to the injury were recorded (@1 fps per channel with a 63X/1.4 NA objective). Axons were previously transduced with lentiviral particles containing GFP or mem-mCherry constructs, and in some experiments incubated also with Fluo-4 AM before the injury. To describe the axonal changes, we developed a workflow applied to all the movies. Images were enhanced to maximize the sharpness of the axonal borders. For mCherry-Mem and GFP, where intensity measurements were not performed, CLAHE filter followed by Top-hat transformations were applied, while for the Fluo-4 AM only linear filters were applied. After the pre-processing step, axonal longitudinal axis was divided into approximately 30 columns with a fixed width (from 1.5 to 3 µm depending on the total axonal length), defining 30 segments. Next, the enhanced image was binarized using Otsu’s threshold that minimizes the weighted within-class variance. For each of the segments a ROI consisting of the two axon shaft borders (i.e. the “upper” and the “lower”) was detected. Thus, all the segments of the axon could be tracked independently during the movie. In this way, for each segment, the distance between one border and the other (thickness) was computed throughout the movie. An axonal swelling is a histopathological hallmark where axons get swollen (enlarged) in a particular region. Thus for detecting AS, we searched for segmental increases in thickness. In a given frame the median thickness of all the segments was calculated resulting in the median thickness of the axon at a given time and set as 100%. From this, the segments that present a thickness of 150% or higher were identified and considered as AS. By iterating this operation in every frame of the movie the development of AS can be tracked in space and time. In the analysis, where number of AS per axons was studied, any AS that was present in the pre-injury stage was subtracted to the whole movie for normalization purposes.
For the Ca2+ levels analysis in conjunction with the membrane changes, Fluo-4 AM intensity levels were assessed in the same ROIs delimited by the membrane marker (mem-mCherry). Thus, for every segment of the axon we attributed an intensity. For analysis of variation of Ca2+ levels through time, intensity levels were normalized to the pre-injury stage levels (mean of first 30 frames). To assess the presence of high Ca2+ levels in the AS we analysed the matrix (N x T) of the axons which consisted of T frames (length of movie in s) and N segments (number of segments dividing the axon). For each coordinate (n,t) we assigned presence/absence of AS, and presence/absence of high Ca2+. AS was present at (n,t) if the thickness value in (n,t) was ≥ 1.5 × the median thickness value of all n at time t. A high Ca2+ level was assigned at (n,t) if the Ca2+ level value in (n,t) was ≥ 1.25 × the mean Ca2+ value of the first 30 t in that n. This Ca2+ normalization was made for establishing the Ca2+ resting levels at the initial period of the acquisition (F0). Swelling detection failed in some cases (n,t) due to loss of focus or errors in the detection algorithm. This led to discontinuity in the tracking of some AS (eg. in a given segment, from 50 to 60 s a swelling was detected but s 61 and 62 was not, followed by 63–80 with swelling). To not overrepresent the AS numbers, we applied a formula to connect these AS. Thus, if in a given segment the AS had a duration of l frames, it merged the following AS only if this was not further than 0.3 × l. After this processing, we performed the analysis of high Ca2+ levels throughout the duration of the AS, with 3 possible categories: high Ca2+ levels were present during all the duration of the swelling, high Ca2+ was not present in any moment of the duration of the swelling, high Ca2+ levels were present during some moments of the AS duration. To assess if high Ca2+ precedes the AS formation, we analyzed the Ca2+ levels of the two frames preceding (2 s) the AS.
Pharmacological blockage of calcium stores
Dose–response experiments were performed to set the drug concentration to use in injury. Neuronal cultures DIV 40 grown in IBIDI chambers were incubated with Fluo-4AM (5 µM in NOMc) for 30 min. Based on previous literature we chose a range of concentrations for every drug and incubated the cultures for 30–60 min in NOMc. After incubation, drugs were maintained in all solutions. Cultures were kept in aCSF and a solution of KCl (50 mM in aCSF) applied. Time lapse movies were recorded at 2 fps for 300 s at 40X/1.3 NA at 480 nm excitation with confocal live module (Zeiss LSM7). First 100 s were used for normalization purposes and were followed by the application of KCl. The mean Fluo-4AM intensity of 15 projections was measured during the treatment, followed by the calculation of the area under the curve.
Axonal injury with pharmacological blockage of Ca2+ stores was performed in microfluidic chambers following the same procedure as detailed before, with minor changes. Briefly, DIV 40 neurons in microfluidic chambers were transduced with mem-mCherry. After 5 days, neurons were incubated with Fluo-4AM for 30 min. Neurons were then incubated with Nifedipine (1 µM), Gadolinium (100 µM), Ryanodine (100 µM), Xestospongin C (10 µM) or CGP317157 (10 µM) in NOMc for 30–60 min. Following incubation compounds were kept in all medias. Incubation was followed by a wash in aCSF and kept in aCSF for imaging. Movies of axons subjected to the injury protocol were recorded (@1 fps per channel with a 63X/1.4 NA objective). Number of AS and Ca2+ intensities during time were calculated as detailed previously. During and post-injury calculations of mean number of AS and mean Ca levels were done by computing the last 30 s of the injury period and the last 30 s of the post-injury period.
Axonal transport
The movement dynamics of APP-GFP, Synaptophysin-GFP and Mito-GFP particles was analyzed using the same protocol. Movies of both control axons and those subjected to injury were acquired at 1fps for Synaptophysin and Mito, and at @2fps for APP, in a confocal microscope equipped with a live module (Zeiss Confocal LSM780, Zeiss Live LSM7) with an immersion oil objective 63x/1.4 NA Plan Apochromat. Time-lapse movies were processed with ImageJ, prior to their analysis in Imaris. Particles were tracked with semi-automated spot tracking algorithm and visualized during the whole period of the movie. The identified movement of each particle during the movie were represented as spots and tracks. First, we choosed the algorithm that would define the movement of our cargoes of interest in the best way. Considering that our cargoes have a complete or almost continuous movement, we applied an Autoregressive Motion algorithm. This semi-automated algorithm requires the input of the following parameters: XY estimated diameter, max distance, and max gap size. Diameter was based on an average empirical value for each specific cargo analysed. For both max distance and max gap size we considered both spatial and temporal resolution before applying any kind of value. Movement along all the axes of the axon was analyzed, but just that along the x axis was taken into account. Among all the detailed statistics obtained from the movies analysis, the most significant one is the spatial displacement of a particle in each frame, which was exported for further computation of different axonal transport parameters. With this value we decided to analyze stationary vs moving tracks, average velocities, and time proportions of anterograde/retrograde/stationary during 3 periods: (1) pre-injury, (2) injury, (3) post-injury. For transport dynamics analysis we first quantified the tracks in stationary or moving status. The stationary tracks were those that didn’t show any movement in the whole period (pre-injury, injury and post-injury) of the movie, presenting an average velocity < 0.010 µm/s. These tracks were excluded from the analysis. For the moving tracks we first set a threshold based on track duration (td = total time during which a particle moves) choosing only those lasting > 10 s. Among the remaining tracks we took for each stage (pre-injury, injury, post-injury) the mean displacement over time. Thus, the movement of the particle for each independent stage was considered anterograde or retrograde if the average velocity was > 0.01 µm/s or < − 0.01 µm/s, respectively, otherwise they were considered stationary.
Structured illumination microscopy
Samples were fixed and immunofluorescence was performed as stated elsewhere. Secondary antibodies used for cytoskeletal proteins were conjugated with Alexa-488 and secondary antibody used for mCherry was conjugated with Alexa-555. Samples were examined in a Zeiss Axioimager Z.1 platform equipped with the Elyra PS.1 super-resolution module for structured illumination (SIM). The high-resolution images were acquired in super-resolution mode using Zeiss Pln Apo 63x/1.4 Oil objective (tot. mag. 1008x) with appropriate immersion oil (Immersol 518F). The SR-SIM setup involved 5 rotations and 5 phases. The software recommended grating was used for each fluorescence channel. For each image, up to 30 Z-stacks (101 nm) were acquired. Image acquisitions and SIM processing were performed in Zeiss Zen 11 software.
Actin and βII-Spectrin periodicity analysis
Z-stacks of actin or βII-Spectrin stained axons and mem-mCherry acquired by SIM were analysed in IMARIS v9. Specific ROIs (shaft, swellings) were cropped and two independent procedures were applied to each channel. The mem-mCherry channel was used to define by auto-thresholding the surface of the AS or the shaft. The channel of the actin or the βII-Spectrin was used to define spots with auto-threshold. All the spots that were outside the volume set with mem-mCherry channel were removed, by using the “spots close to surface” command. The distance between the spots and their closest neighbour was measured with the “spot to spot closest distance” command. Measurements and quantitative data was exported and processed in Excel or Prism.
Validation of the method was conducted by comparing the IMARIS closest neighbour results with intensity plots that result from the trace of a longitudinal line to the mayor axis of the axon shaft [75]. For the latter, the distances between peaks of the intensity profile were plotted as a frequency distribution. This frequency distribution was similar to the frequency distribution obtained by the closest neighbour method (Additional file 13: Fig. S4B).
Scanning electron microscopy imaging
Following injury, chambers were washed with PBS and all the wells filled with fixation solution (Glutaraldehyde 6%, Sodium Cacodylate Buffer 0.2 M pH 7.35, 1:1) for 1 h at RT. The chambers were washed with Cacodylate Buffer, and dehydrated in a series of increasing concentrations of ethanol (2 × 10 min 50°, 2 × 20 min 80°, 2 × 20 min 100° EtOH). Chambers were separated from the coverslip by peeling off and coverslips cut with a diamond pen to isolate the region of interest. Samples were left to dry and coated with an automatic sputter coater (JEOL JFC-1300) for 15 s at 40 mA for enhancing the contrast. Imaging was performed in a JEOL JCM-6000 scanning electron microscope at high vacuum mode, secondary electron image and accelerating voltage of 10 kV.
AS lengths and widths were measured in SEM images of control and injured axons with ImageJ. Inter-swelling distances were measured between the end of AS to the beginning of the adjacent one. Percentage of number of AS per axons were fitted by using a Gaussian model with least squares regression on Prism 9. Relative frequencies of inter-swelling distances were modelled as a Gaussian mixture with parameters determined by the maximum-likelihood estimate (MLE) via the expectation–maximization (EM) algorithm as implemented in the R-package mclust.
Transmission electron microscopy
Following injury, chambers were washed with PBS and fixed with 2.5% glutaraldehyde and 2% PFA in 0.1 M phosphate buffer (pH 7.4) for 2 h at RT. After fixation, chambers were gently peeled off from the coverslips. Contrast enhancement with 1% OsO4 in phosphate buffer for 30 min and 1% uranyl acetate in 50% ethanol/water for 30 min was performed. Samples were then dehydrated in a graded ethanol series and embedded in Durcupan. Ultrathin sections (70 nm) were sliced using a microtome (EM UC7 Ultramicrotome, Leica Microsystems) and mounted on formvar coated copper slot grids. Images were acquired with a transmission electron microscope (Tecnai 10, FEI; operated at 80 kV, equipped with OSIS Megaview III camera).
Antibodies and Reagents
Antibodies, vectors, reagents, kits and software can be found in Additional file 20.
Quantification and statistical analysis
Graphpad Prism 9.3.0 was used for all statistical analyses. Statistical tests are as described in text and figure legends.